Software Engineer, Foundation Industries

Most software lives entirely inside a screen. At Foundation Industries, the software ends at a CNC machine cutting metal. My work here is full-stack platform engineering for manufacturing operations — the systems that connect a customer's CAD file to a quote, an engineering review, a production plan, a machine on the floor, and a finished part out the door.

The problem space

Manufacturing quoting isn't a checkout form. A single order has to reconcile incomplete customer inputs, part geometry, materials, tolerances, quantities, delivery expectations, machine constraints, and internal engineering judgment — and stay consistent across all of them. The interesting engineering is in turning that ambiguity into a dependable workflow that a customer, an engineer, and a machinist can all trust.

What I've worked on

CAD & engineering review

I contributed to an interactive 3D file-review experience: direct geometry interaction, feature selection, and design-for-manufacturability (DFM) annotations at the part level. The real story isn't "I built a 3D viewer" — it's that this changed how engineering intent and manufacturability feedback move through the order pipeline, replacing older GLB-based annotation flows with direct feature selection.

Machine monitoring

Collecting data is easy; making it operationally useful is the hard part. I helped implement machine-data infrastructure at a safe level of abstraction:

1. 01CNC machine (Haas) emits MTConnect-style data.
2. 02A Raspberry Pi collector/proxy polls and forwards it.
3. 03Structured ingestion normalizes the stream.
4. 04PostgreSQL / Supabase stores it.
5. 05Dashboards, KPI reporting, and scheduling signals consume it.

Those dashboards surface spindle time, runtime, utilization, program state, and available-production-slot calculations — the numbers that actually drive scheduling and capacity decisions, not just a wall of telemetry.

Quoting, scheduling & operations

Across the platform I've worked on quote creation and editing, delivery tiers and pricing consistency, purchase orders and paid-order locking, production-slot and machine-capacity visibility, order milestones and ownership, and the operational glue around them (administrative order management, audit history, and communications workflows). The through-line: software that sits close to a physical process and has to respect its constraints.

Reliability & platform integrity

My work here isn't only feature UI. It's also included route protection and permissions, secret-protected scheduled jobs, Stripe webhook idempotency, audit ledgers, and Playwright / Vitest end-to-end testing with CI quality checks and test-database hygiene. Building software that touches real orders and real money means treating reliability as a feature.

What this role is teaching me

How to build software around physical operations — manufacturing constraints, machine data, payments, and the humans (machinists, engineers, customers) on either end of a workflow. It's the clearest example I have of going deep enough to understand a real, messy domain and broad enough to ship across the whole stack.

Status

Current role — Software Engineer, Redwood City, CA (full-time for Summer 2026), at Foundation Industries (Sava Robotics, Inc., DBA Foundation Industries).